An Investigation Into the Impact of Impedance Measurement Parameters on the Limit of Detection of QCM-D Using Machine Learning Model Chaining

Impedance spectroscopy is an appropriate measurement method for quartz crystal microbalance with energy dissipation (QCM-D) monitoring, especially for machine learning (ML) applications, given the vast amount of information it can provide. When quartz crystal microbalance (QCM) is used in a liquid medium for biosensing, it responds to mass change, and the viscoelastic properties of both the medium and the film are deposited on the electrode surface. It has been previously observed that the limit of detection (LOD) experiments employing QCM may be increased by at least 12-fold, by an application of ML-assisted optimization of impedance measurement parameters while enabling a reduction of the number of experiments by tenfold. In this study, ML methodologies are employed to quantify how a selection of such measurement parameters is possible and affects the calculated viscoelastic parameters of the bulk fluid and thickness along with the viscosity of bovine serum albumin (BSA) thin films adsorbed on gold electrodes. Results indicate that the LOD for bulk fluid viscosity and thickness of BSA thin films can vary up to sixfold and threefold, respectively, depending on the chosen measurement parameters. By implementing this ML framework, viscoelastic modeling accuracy in complex media and thin-film applications can be significantly improved through impedance spectroscopy, thus resulting in an increased overall sensitivity in QCM biosensing.